Design Discussion: High Performance CNC Spindle

macardoso

H-M Supporter - Silver Member
H-M Supporter - Silver Member
Joined
Mar 26, 2018
Messages
2,724
I have been doing research on building a high performance spindle for a medium to large size benchtop CNC mill for a few months. I recently found out that @shooter123456 was also working on similar goals and I wanted to create this post to collect information and share ideas. I'm sure he and I have slightly different end goals in mind, so I will try to keep this generic.

Background: Anyone who has converted or built a benchtop CNC (up to RF45/PM-940 size) is likely aware of the extreme limitations of spindle options out there. In fact, the spindle is almost always the limiting factor in custom built designs. There are a couple of categories that we can group the available spindles into:
  • Stock Spindle: These are the machine running the original spindles that came with the machine. While this doesn't add much to the cost, these spindles are often limited in speed and torque due to a gear head design and brushed or brushless DC motors. Speed control from the CNC controller may be difficult without special interfaces.
  • High Frequency Spindle: There are a ton of router type machines running water or air cooled high frequency AC spindle motors coming from China at relatively low prices. For general machining work, the 18-24k rpm spindle is too fast and the extremely limited torque and rigidity limits their application to light cuts with small tools.
  • AC Induction - Belt Drive: Many hobbyists opt for an open loop AC induction motor run on a VFD. The cost of this setup is often significantly more expensive than the aforementioned options, however it offers great torque and control over the motor speed. Hobbyists are usually limited by the size and weight of the motor, as well as the cost. Speed is limited without special considerations to 7200rpm (2x motor nameplate) and torque falls off rapidly above 3600 rpm. Rigid tapping is possible with special considerations.
  • Servo Spindle - Belt Drive: Typically seen on the high end hobby builds, a servo spindle offers a flat torque profile across the entire speed range and precise control of velocity and position. Rigid tapping is easily accomplished with this spindle configuration. Hobbyists are often limited by the cost of these systems and the maximum speed of the lower cost models (3000rpm). Servos rated for 5000 rpm are available up to 2.5HP but may be difficult to find.
With these options in mind, we look at some features which an ideal spindle (again, for a high end hobby class machine) would have:
  1. 8000-12000 rpm: Higher spindle speeds offer cutting performance for tools 3/8" and under.
  2. 3HP: Roughly the maximum single motor rating before 3 phase power is required.
  3. Substantial low end torque (>30Nm): Allows use of larger diameter tools, facemills, and improved performance in steel.
  4. <75 dBA Noise: Comfortable to work around.
  5. Thermally stable: <130*F at 100% duty cycle.
  6. Quick Change/ATC Compatible: Pneumatic/Hydraulic tool release
  7. Low TIR: .0002-.0001 TIR at spindle taper
  8. Closed Loop: Encoder Feedback of spindle position for speed control and rigid tapping
  9. Common Taper: R8/BT/CAT taper, minimal tool change height is ideal due to limited travel in hobby machines
  10. High Spindle Rigidity: Thick spindle walls and large bearings for improved cutting performance
  11. Tool Compatibility: Tooling system equipped to hand tools up to and including 1/2" shank. Options available for larger tools if needed.
  12. Cost: Cost should not exceed $1-2000 over the purchase price of the machine.
In upcoming posts, I'll try to dive into research I've done on each of these topics and how we might achieve them. Please join in and share your ideas or let me know if any of the requirements should be tweaked.

-Mike
 
Discussion about target spindle speed

Many of the benchtop machines which hobbyists target for CNC conversions come from the factory with geared head spindles with a top speed of 2500-3000rpm. These sppeds are suitable for manual machining with HSS and drilling, but is far from ideal for dynamic (trochoidal) machining with smaller cutters. A machine the size of a G0704/PM25 seems to be best suited for using 3/8" endmills in aluminum and 1/4" in steel (full flute length 10-40% engagement), while a machine the size of a RF45/PM940 would be suited for 1/2" endmills in aluminum and 3/8" in steel. Larger cutters may be used but rigidity is very limiting. Likewise, small cutters are acceptable, but productivity is extremely limited by spindle speed.

Using modern carbide tooling (seemingly commonplace among serious hobby CNC users) the allowable spindle speeds for tooling is very high. Far beyond the 3000rpm limitation of the stock spindles. It is for this reason that we would want to consider a 8-12k rpm spindle, especially for aluminum.

Having recently installed a 5000rpm servo spindle on my G0704, I can confidently say that I still feel limited by my top speed, especially when using tools under 1/4". My 3/8" aluminum specific carbide endmills recommend 10-12k rpm in their speeds and feeds. A higher speed would go a long way in making these machine more productive in finishing operations with small tools.

Spindle Options

Unfortunately reaching these higher speeds presents a real design challenge that goes against many of the other design criteria listed in the first post. The first challenge is that there are not many motors capable of providing 8-12k rpm directly. Gearing up the speeds is feasible but presents a large number of challenges specifically with noise, efficiency, and torque drop. Standard gearing available at the hobby level cannot cope with the increased speed and stresses. Belts perform poorly when the driven pulley is much smaller than the driver. Direct drive attachments limit access to the spindle for drawbar usage. Finally, built in spindle motors require extensive machining and engineering, and I have never once seen one used in a hobby machine.

From my design analysis, belted connections to a motor with a very high name plate speed is the best option, reducing the ratio which must be introduced to reach the final target speed. The easiest motor to use would be an inverter duty induction motor, giving a running speed of 7200rpm. This configuration would necessitate a 1:1.38 gearing ratio to hit the 10k rpm mark.

A better solution would be to use a purpose built spindle motor (such a a Fanuc a2, A06B spindle motor) which comes ready to run at 8-10k rpm. In real Fanuc CNC machining centers, these motors are belted to the spindle and provide power and torque very effectively. I have been scouring for used ones and have seen them listed for $500-1000. This motor could be geared 1:1 or a small increase ratio which would be very acceptable. Purpose built spindle motors have a constant blower fan to maintain cooling in low rpm high torque applications.

If servo motors are preferred there are a few high speed options available in the 6000-8000 rpm range from Siemens, Kollmorgan, and perhaps others. These tend to be lower power, low inertia units which may not be favorable for a spindle motor with medium inertia.

Transmission

As suggested above, it seems that at these high speeds, you are limited to direct drive or carefully engineered belt drive transmission systems. Gearing is used in industry, but requires hardened and ground gears in special high speed gearboxes. It is my opinion that this is out of reach for the hobbyist.

From my experience, timing belts are not necessary when the belt transmission is appropriately designed, and at higher speeds, they are a significant source of running noise. Air is sucked into the teeth and tightly compressed as the tooth rotates around each pulley. The compressed air then snaps out the other side creating a lot of noise. My current CNC timing belt spindle design operates comfortably at 5000rpm, but would not be suitable for higher speed applications.

Poly-Vee belts are marketed as the preferred solution for high speed applications and have been proven throughout the industry, both in machine tool and automotive applications. The thin profile of the belt reduces the size requirements of the pulleys.
 
Last edited:
Discussion about HP

This is a fairly easy one. We would like to have the most horsepower available at the tool (within reason). The machines this spindle is being designed for will have significant rigidity limitations so it is unlikely that you would be able to take advantage of a 5 HP spindle. Furthermore, most VFD manufacturers switch to 3 phase input above 3HP, marking it as a good upper bound.

My current G0704 has a 2.25HP spindle and I am hard pressed to use more than 40-50% of its capacity while milling, due to rigidity limitations. I am able to get close to 100% in drilling operations.

If a hobby machine were suitably constructed to handle the cutting forces, and the user was willing to install a rotary phase converter (or oversize their VFD for single phase derating) then a larger spindle motor could be considered.
 
Macardoso, I appreciate your thoughtful discuss on these topics. I have PM45-CNC and have often thought about what it would take to increase the maximum RPM.
 
Discussion about low end torque

In a milling spindle, there is a unique range of applications, from large drilling and facemilling, to micro tooling. The high spindle speeds typically do not require substantial torque, however the tools running at lower speeds do. A motor which has been geared up for high speed use will perform quite poorly when using larger tools in steel or using facemills.

My current design, for reference, uses a 1:1 servo motor with 11Nm of torque at 100% continuous, with momentary spikes to 300%. This is plenty sufficient for milling aluminum and steel at 5000rpm, however it is quickly used up when using larger cutters (>1/2") or facemills. I would suggest a minimum of 30Nm for acceptable performance in the <1500rpm speed range.

It becomes quickly apparent that this is likely not able to be achieved with a single gearing ratio, and a belt change or back gear would need to be employed.

Transmission

I haven't done the most research into the design, but Fadal, a CNC machine manufacturer, has a really neat pneumatic belt change system. It has multiple belts on a step pulley system, but only 1 can be in contact with the pulley at a time. They are switch by a pneumatic system which engages or disengages an idler pulley. Perhaps this system could be integrated into out high performance spindle to give multiple speed and torque ranges.

1579028794195.png

The alternative option is a classic step pulley which would require the user to stop the machine and change the belt. I believe this is what Tormach went with on their new 1100MX mill with a 10k spindle.
 
Noise

I may have more sensitivity to this issue than others, but my hobby shop is in my basement of a 120 year old house. The walls and floors are very thin and my wife doesn't appreciate a lot of noise when I'm working in the shop. In addition, I find the machine much more pleasant to use if I can be around it comfortably without damaging my hearing.

The design limit I set when building my current CNC was 75dBA motor idle noise at 5000rpm (doesn't count the cutting sounds of endmill on workpiece). To accomplish this, I had to carefully select a profile of timing belt for high speed applications and totally enclose the transmission in a thick walled airtight housing. The final sound level ended up at 68dBA as measured by my ultra accurate iPhone app at 1' from the front of the machine

I believe that designing this high performance spindle in a similar way from the beginning would be important. The higher speeds present a ever greater challenge than I faced. Here are some design factors I think will play the most into the audible noise:
  • Quality spindle motor with good bearings. Should be very quiet during a bench test.
  • Smooth running and balanced spindle with correctly installed spindle bearings.
  • Selection of a non toothed Poly-Vee belt. No cogging noise like a timing belt.
  • Sealed transmission enclosure, limits emitted noise, safety. Downside, difficult to access for manual belt changes. Great for automatic belt change
  • Low TIR (<0.001") of driver and driven pulleys
  • Minimal unsupported span of belt between motor and spindle
  • Easily adjustable spindle belt tension. Adjustment screw would be great here
At the end of the day, the spindle will still be noisy, but careful design will minimize this. A built in blower fan will add a good chunk of noise that we cannot control.
 
Thermally stable

I have not done much/any research into this, however the spindle bearings on my 5000rpm spindle do tend to get to 120/130*F during prolonged running at full speed. The bearings are safe to around 250*F, however this is way too hot and the grease will drip out of the bearings long before this.

Carefully set preload of the spindle bearings can minimize heat, however at 8-12k rpm, the heat generated may be more than passive cooling can handle. A simple water cooling jacket around the spindle bearings and perhaps the head casting itself would do a lot to help stabilize the spindle temperature, however it does add a lot of complexity.

Thermal management is also important for accuracy. On an RF45/PM945 machine, starting the spindle from room temp (70*F) and allowing it to reach 130*F (assumption, constant temperature all the way to the column) would cause the tool to move 0.0029" in the negative Y direction. No hobby CNC control that I know how has thermal compensation in it, so this error will throw your parts out of tolerance.
 
Quick Change/ATC Compatible

Since most serious hobby CNC users will be using preset tools and maybe even an ATC, the spindle must be able to automatically clamp and unclamp the tool. The exact mechanism will vary depending on if you are using a BT taper with a pull stud or an R8 taper and Tormach Tooling System.

The tool retention force must be sufficient to handle the cutting forces of this spindle, and the throw of the drawbar must be great enough to fully unclamp the tool.

My current R8/TTS system retains the tool with 1800lbs of force, however I have had a couple incidents of tool pullout, so I would consider a system closer to 3000lbs. BT30/40 tapers have different requirements which I have not researched.

Typically the hydraulic or pneumatic cylinder is located directly above the spindle, so having the spindle motor off to the side with a belt allows access for this.

The mechanism should not place the compression forces across the spindle bearings.

EDIT: A properly designed tool clamping mechanism should have feedback to ensure tool clamping and separately feedback to ensure the cylinder is not in contact with the spindle during rotation.
 
FWIW, I measured the sound levels from my Tormach 770. At maximum speed the motor generates 90-92 dB @ 3ft. With the spindle in the high range, it goes up to around 95 dB while the spindle in the low range, it is around 92 dB. I suspect that most of the noise is due to the integral fan in the motor. There have been a number of mods to the Tormach's removing the internal fan and replacing with an external fan. While this was done for improved cooling at lower speeds, it would cut the noise quite a bit.
 
FWIW, I measured the sound levels from my Tormach 770. At maximum speed the motor generates 90-92 dB @ 3ft. With the spindle in the high range, it goes up to around 95 dB while the spindle in the low range, it is around 92 dB. I suspect that most of the noise is due to the integral fan in the motor. There have been a number of mods to the Tormach's removing the internal fan and replacing with an external fan. While this was done for improved cooling at lower speeds, it would cut the noise quite a bit.

Yikes! That is a lot louder than I imagined it would be. Thanks for sharing! Maybe my suggested requirement is too tight then?
 
Back
Top